Figshare seems to allow sharing any size of data publicly and assigns a DOI for citation. Content is hosted on Amazon S3 and archived at different research libraries. The combination of a DOI and third party storage should avoid the URL-rot that befalls many published URLs pointing to lab/university web sites for code or data. One possibility would be to create a virtual machine image containing all code and data needed to re-capitulate a published analysis. Or one could use the (linux only) CDE tool i posted about earlier.